--- title: "When Large Models Enter the Physical World: Xiaomi Aims to Add “Memory” to Smart Homes" type: "News" locale: "en" url: "https://longbridge.com/en/news/290203821.md" description: "On June 18, Xiaomi released Xiaomi Miloco 2.0, an open-source AI solution for whole-home intelligence. Centered on its self-developed MiMo large model, the solution introduces a Home Memory AI system for the first time, capable of identifying family members, their preferences, and daily routines to build long-term memory. Leveraging multimodal perception through Mi Home cameras and other devices, it upgrades from “rule-based control” to understanding living habits, supporting Agent plugin integration across multiple operating systems" datetime: "2026-06-18T13:04:50.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/290203821.md) - [en](https://longbridge.com/en/news/290203821.md) - [zh-HK](https://longbridge.com/zh-HK/news/290203821.md) --- # When Large Models Enter the Physical World: Xiaomi Aims to Add “Memory” to Smart Homes Smart homes are evolving from “understanding a command” to “remembering a home.” On June 18, Xiaomi officially launched the “Whole-Home Intelligence AI Open-Source Solution” Xiaomi Miloco 2.0. Built around Xiaomi’s self-developed MiMo large model as its intelligent core, Miloco 2.0 upgrades interaction methods, product features, and the memory system based on last year’s Miloco 1.0. Miloco 2.0 primarily integrates with OpenClaw via Agent plugins, supporting macOS, Linux, and Windows systems. **The most noteworthy change is Xiaomi’s introduction of a Home Memory AI system for the first time.** Previously, the core capability of smart homes was mainly device linkage. Users would set rules in advance, and the system would execute actions accordingly—for example, turning on the living room lights at 7 p.m., activating the air conditioner when the temperature exceeds 28°C, or responding to a voice command like “turn off the lights.” However, such smart home systems are essentially still “rule-based controls.” They can execute explicit instructions but struggle to understand the long-term living habits formed by a household, let alone continuously adjust services based on the preferences of different family members. Miloco 2.0 aims to fill this gap. According to Xiaomi, Miloco 2.0 can remember the identities, preferences, schedules, and habits of different family members, accumulating long-term memory on a “household” basis. It organizes observed patterns every early morning, consolidating recurring behaviors into long-term profiles. For instance, if the system notices a family member returning home late, it may proactively offer care related to overtime work. To support this home memory, Miloco 2.0 uses Mi Home cameras as a multimodal perception entry point, combining microphones, Mi Home devices, and large model capabilities to continuously understand home scenarios. The system can determine family members’ identities by synthesizing information such as facial features and body shapes. If identification is temporarily unsuccessful, the individual is placed in a “stranger pool” pending user confirmation before registration is completed. This addresses a core issue arising from the integration of AI large models into home scenarios: without long-term memory, large models remain limited to one-off Q&A and single-command execution, which is no different from the previous “rule-based control” in smart homes. From a broader technological perspective, the significance of Miloco 2.0 lies in providing a sample case for observing how large models enter the physical world. In the past, large models primarily operated within chat boxes, search bars, and office software, with interactions mostly confined to screens. However, the home environment is a real physical space. For AI to be effective here, it must continuously perceive the environment, identify family members, understand behavioral changes, and dispatch real devices when necessary. In this sense, home memory is a crucial capability that large models must develop as they enter the physical world. Risk Warning and Disclaimer The market involves risks; investment should be approached with caution. This article does not constitute personal investment advice, nor does it consider the specific investment objectives, financial status, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investment decisions made based on this content are the sole responsibility of the investor. ### Related Stocks - [XIACY.US](https://longbridge.com/en/quote/XIACY.US.md) - [01810.HK](https://longbridge.com/en/quote/01810.HK.md) - [81810.HK](https://longbridge.com/en/quote/81810.HK.md) - [HXXD.SG](https://longbridge.com/en/quote/HXXD.SG.md) ## Related News & Research - [Alibaba Launches Robotics AI Models as It Ramps Up Physical AI Push](https://longbridge.com/en/news/289864981.md) - [MoneyMasters Podcast: Armani on Why the AI Boom Is Different](https://longbridge.com/en/news/290232066.md) - [Nocera Strengthens AI Growth Strategy with Strategic Investment in CampaignPulse.ai](https://longbridge.com/en/news/289958177.md) - [Zhipu AI’s stock rockets after Chinese firm makes GLM-5.2 open source](https://longbridge.com/en/news/289742970.md) - [08:47 ETdeepset Joins HPE's Unleash AI Program to Accelerate Sovereign Agentic AI](https://longbridge.com/en/news/290200766.md)